Improved Radon transforms for filtering of coherent noise

Shauna Kaye Oppert, R. James Brown

Radon transforms rely on the ability to predict the moveout of coherent events. Most algorithms assume parabolic or hyperbolic moveout, a characteristic that many reflections do not adhere to. Standard parabolic and hyperbolic Radon transforms typically involve smearing of reflections across Radon space, which reduces the effectiveness of coherent-noise suppression. We present a method developed to specifically remove reflections having nonhyperbolic moveout. The shifted-hyperbolic and anisotropic Radon transforms employ a curve-fitting technique to allow for flexibility in predicting the true moveout of specific reflections. In addition, approximations to the damping factors used in the low- and high-resolution Radon algorithms are presented. These alternate parameters are feasibly employed and improve the efficiency of the algorithms.